A blood-based biomarker panel to risk-stratify mild traumatic brain injury.

Abstract

Mild traumatic brain injury (TBI) accounts for the vast majority of the nearly two million brain injuries suffered in the United States each year. Mild TBI is commonly classified as complicated (radiographic evidence of intracranial injury) or uncomplicated (radiographically negative). Such a distinction is important because it helps to determine the need for further neuroimaging, potential admission, or neurosurgical intervention. Unfortunately, imaging modalities such as computed tomography (CT) and magnetic resonance imaging (MRI) are costly and not without some risk. The purpose of this study was to screen 87 serum biomarkers to identify a select panel of biomarkers that would predict the presence of intracranial injury as determined by initial brain CT. Serum was collected from 110 patients who sustained a mild TBI within 24 hours of blood draw. Two models were created. In the broad inclusive model, 72kDa type IV collagenase (MMP-2), C-reactive protein (CRP), creatine kinase B type (CKBB), fatty acid binding protein-heart (hFABP), granulocyte-macrophage colony-stimulating factor (GM-CSF) and malondialdehyde modified low density lipoprotein (MDA-LDL) significantly predicted injury visualized on CT, yielding an overall c-statistic of 0.975 and a negative predictive value (NPV) of 98.6. In the parsimonious model, MMP-2, CRP, and CKBB type significantly predicted injury visualized on CT, yielding an overall c-statistic of 0.964 and a negative predictive value (NPV) of 97.2. These results suggest that a serum based biomarker panel can accurately differentiate patients with complicated mild TBI from those with uncomplicated mild TBI. Such a panel could be useful to guide early triage decisions, including the need for further evaluation or admission, especially in those environments in which resources are limited.